The common carp (Cyprinus carpio) is an important freshwater species for aquaculture and has been domesticated for ca. 4000 years [1]. Annual production of common carp has been increasing for 30 years, and recent production reached 3.3 million tons [2]. Numerous microsatellite markers [3–5] have been developed in common carp in response to the demands of genetic and breeding studies. Compared with microsatellites, single nucleotide polymorphisms (SNPs) are more promising molecular markers in genetics, genomics and aquaculture. Because of their high abundance in the genome, hereditable stability and allele portability, and the possibility for high-throughput analysis, they may replace traditional molecular markers in genetic studies [6,7]. EST-database mining is an efficient way to obtain SNP markers and was first applied for SNP discovery in human sequences [8], EST-SNPs were also obtained in many fish species [9–12], while to date only limited EST-SNPs have been characterized [13] in common carp. In this study, discovery of a set of EST-SNPs from online common carp EST databases and validation of selected EST-SNPs in the Chinese common carp were carried out, with the aims of (1) developing novel SNP resources for future studies of genetics and breeding in common carp; (2) validating and characterizing SNPs in common carp with different genetic backgrounds for the comparison of genetic variations.

The alignment of 14,066 online common carp ESTs resulted in the identification of 8862 unigenes (6727 singletons and 2135 contigs). The number of EST sequences per contig ranged from two to 60 (mean 3.4). Based on searches of the 2135 contigs, a total of 49,476 variations were detected, including 40,749 nucleotide substitutions and 8509 indels. After the additional filtering criterion was applied, a total of 1487 putative SNPs were obtained from 303 contigs, with an average frequency of one SNP per 173 bp of contig sequence, which was greater than frequencies reported in Atlantic salmon (one SNP per 614 bp) [10], brown trout (one SNP per 463 bp) [14] and Pacific salmon (one SNP per 239 bp) [15], but close to the frequency reported in chum salmon (one SNP per 175 bp) [15]. The distribution of SNPs in these EST contigs ranged from one to 32 (Figure 1). Of the unigenes described in this study, 4127 (46.7%) were annotated, and the annotations for the contigs containing valid SNPs in the Chinese common carp are listed in Table 1.

Of the 81 pairs of primers, 48 produced PCR products. Among these, 30 amplified products were larger than the expected size, indicating the existence of introns (Tables 1,S1). A total length of ca. 27,010 bp was amplified and sequenced, of which 13,213 bp were exons and 13,797 bp were introns. Finally, 121 SNPs were identified from 13,213 bp of exons, and 26 (21.5%) of these putative SNPs were verified in a Chinese common carp population. On the other hand, a large number of new SNPs, which were not detected in the SNP discovery by data mining from the online EST sequences, were identified during the validation. These new SNPs included 202 SNPs and 10 indels in exons, and 306 SNPs and 58 indels in introns (Table 1, Figure 2, Supplement Material 1). The distribution of SNPs in exon sequences ranged from one to 34 (Figure 1). In addition, the average rates of occurrence were about one SNP per 58 bp in the exon sequences, one SNP per 45 bp in the intron sequences, and one SNP per 53 bp in the complete genomic sequences. These occurrence rate are much higher than the rates of putative SNPs discovered from the online EST databases in most fish species, but quite close to that of the hybridized catfish (1.32 SNP per 100 bp) [16]. In the present study, although the validation rate of predicted SNPs from online ESTs of European originated common carp was low, many new SNPs were found during the validation by sequencing the PCR products from the Chinese carp. These findings may suggest that different common carp breeds or geographic populations have different levels of genetic divergence and population structures, which are indicated by the gain or loss of SNP loci at a great level. However, these results should be interpreted with caution since the SNP frequency in discovery could be affected by many factors, such as sequencing depth, number of samples and population variation. Another factor that may be involved in this phenomenon is the higher rates of genome duplications in the common carp as indicated in other tetraploid fish such as some salmonids [15,17,18].

Among the 228 exon SNPs, 145 were transitions (81 C/T, 64 A/G) and 82 were transversions (27 A/C, 19 C/G, 18 A/T, 18 G/T), while in intron SNPs there were 154 transitions and 149 transversions (Table 2). Two SNPs had both transition and transversion (C/G/T and A/G/T), this type of SNP has also been detected in Atlantic cod [11]. The ratio of transitions to transversions (t_s/t_v) was 1.77 for exon SNPs, 1.05 for intron SNPs and 1.31 for SNPs in complete genomic sequences. These ratios are similar to those reported in turbot (1.885) [12], chinook salmon (1.49) [15], gilthead seabream (1.375) [9] and zebrafish (1.20) [19]. However, in other fishes, such as chum salmon (t_s/t_v = 0.95) and sockeye salmon (t_s/t_v = 0.98) [15], the t_s/t_v ratios were significantly lower, and only close to the ratio estimated from the intron regions of common carp ESTs (1.05) in this study. This discrepancy in t_s/t_v ratios may suggest biased codon usage or substitution rate because fishes from different phylogenetic units may be subject to different selection pressures. Comparison of exon and intron SNPs revealed extreme differences in almost all aspects, including SNP frequency, t_s/t_v value, and commonest and rarest SNP types (Figure 2, Table 2). This may be explained by dissimilar natural selection pressures on exons and introns. In exons, most destructive mutations, which cause loss of protein function, vanish with less fit individuals, while beneficial mutations are retained and accumulate during evolution. Conversely, evolutionary constraints on introns are relatively weak, and more variations could be retained [20,21].

Twenty-three SNPs were chosen to test the polymorphisms in 38 unrelated individuals from a Yangtze River common carp population. All these loci were found to be polymorphic, with frequencies of minor allele ranging from 0.053 to 0.368 (mean 0.19; Table 3), observed heterozygosity from 0.053 to 0.526 (mean 0.262; Table 3), and polymorphism information content (PIC) from 0.095 to 0.357 (mean 0.246; Table 3). In the exact test for HWE, 21 SNPs were in HWE (p > 0.05), and 2 SNPs deviated significantly from HWE after Bonferroni corrections (p < 0.00217). Linkage disequilibrium (LD) was detected between 19 pairs of the SNP loci after Bonferroni corrections (p < 0.00020).

A total 1487 putative SNPs were identified by mining from online common carp EST sequences. Approximately 320 of the putative SNPs are expected to be true in the Chinese common carp as estimated based on the validation rate of 21.5% in this study. In spite of the low validation rate, large numbers of new SNPs were identified in the Chinese common carp. Of the 23 SNPs tested, all loci were polymorphic in a Yangtze River population with moderate diversity. These results indicate that the occurrence of SNPs varies significantly between European and Chinese populations, and these characterized SNPs are valuable resources for population genetics, high-resolution genetic maps, QTL (quantitative trait locus) identification, and maker assisted breeding in the common carp.